Journal of Sol-Gel Science and Technology 18, 269–283, 2000c© 2000 Kluwer Academic Publishers. Manufactured in The Netherlands.

Cells in Sol-Gels I: A Cytocompatible Route for the Productionof Macroporous Silica Gels

JOHN F.T. CONROY, MARY E. POWER, JASON MARTIN, BRIAN EARP, BOUVARD HOSTICKA,CHARLES E. DAITCH AND PAMELA M. NORRIS∗

Department of Mechanical and Aerospace Engineering, University of Virginia, Charlottesville, VA 22903, USApamela@virginia.edu

Received January 7, 2000; Accepted June 5, 2000

Abstract. A novel, high hydrolysis ratio sol-gel route for the biocompatible production of macroporous silica gelsis presented. This route exploits the two step nature of the gelation reaction to remove undesired alcohol by-productsfrom an acidic aqueous sol prior to gelation. These alcohol-free sols will gel when the pH is raised to the physiologicrange in a two-step, acid/base catalyzed process. Furthermore, monolithic macroporous samples can be producedin a controlled manner by introducing water-soluble organic polymers into the sol.

Keywords: bacteria, immobilization, sol-gel production, macroporous

Introduction

Sol-gel-derived silica holds much promise as a cyto-compatible scaffold for the immobilization of cells ina variety of applications. There are several favorablecharacteristics of sol-gel-derived silica as cell immo-bilization matrices, including a low temperature pro-duction route, chemical-, temperature-, and radiation-stability, a very high surface area and porosity, easeof functionalization, mechanical rigidity (no swelling),and tunable properties and microstructures. Further-more, the biocompatibility of both sol-gel-derived andnon-sol-gel-derived ceramic materials has already beenextensively investigated [1–6].

Despite this promise, limited progress in the use ofsol-gel derived silica as a cell immobilization matrixhas been made. Common sol-gel production methodsare too cytotoxic at the time of gelation for extensiveuse in the immobilization of cells. Furthermore, macro-porous samples amenable to colonization are difficultto obtain and may require the use of toxic chemicals.In early immobilization procedures, sol-gel-derived

∗To whom all correspondence should be addressed.

immobilization matrices were either rinsed and sub-sequently loaded with a cytocompatible liquid phaseafter gelation but before loading with microorganisms[7] or they were limited to robust species that could sur-vive the relatively harsh conditions at gelation [8–10].

This paper presents a sol-gel production route wheregelation occurs almost entirely in water. This produc-tion route is capable of reproducibly yielding mono-lithic micro- and macroporous samples with surfaceareas in excess of 1000 cm2/gram. Microstructure as afunction of temperature at gelation, ratio and nature ofthe reactants, and aging time of the prehydrolyzed solis investigated.

Background

The immobilization of cells within sol-gel-derived sil-ica matrices is an idea that has held much promise forseveral years. In the late 1970’s, Hino et al. describeda two-step synthetic route for immobilizing micro-organisms within complex gels containing significantorganic and sol-gel-derived silica fractions [11]. Theweight percent of silica in these gels was relatively low

270 Conroy et al.

and the gels themselves often displayed behavior char-acteristic of organic polymer gels, such as significantswelling in different solvents and solubility in water.In the 1980’s, Carturan et al. immobilizedS. cerevisiaein a multilayered sol-gel-derived thin film [8]. In the1990’s, Uo et al. pre-gelled macroporous silica matricesin solutions containing water-soluble polymers, rinsedthe gelation solution from the gels, and then loaded theresulting macroporous samples withSaccharomycescerevisiaeand showed germination [7]. Pope et al. useda polymer-free, two-step synthetic route to immobilizebothS. cerevisiae[9, 12–14] and pancreatic islet cells[15, 16] by adding the cells to a prehydrolyzed silica so-lution prior to gelation. He has also developed a methodfor making cell-containing, sol-gel-derived silica mi-crospheres [17]. In a similar production route, Livageet al. mixedLeishmania donovani infantuminto pre-hydrolyzed methoxysilicate sols that were later gelledand optically probed [18]. Al-Saraj et al. immobilizedS. cerevisiaein silica gels derived from tetraethoxysili-cate and examined the bioaccumulation of heavy metals[10]. Branyik et al. encapsulated both yeast (Candidatropicalis) and bacteria (Pseudomonas) inside similarsilica gels and investigated their use in bioremediation[19]. They found that the sol-gel encapsulation proce-dure was relatively benign, but that the organisms wereso constrained by the silica matrix that they were un-able to colonize the scaffold. Rietti-Shati et al. also im-mobilized a strain ofPseudomonaswithin silica gelsusing both pure sol-gel-derived silica and combinedsilica/Ca-alginate system. They obtained best resultswith the pure sol-gel-derived silica system. An alter-native to encapsulation in a liquid sol is termed Biosiland involves cell immobilization upon fiber supportsfollowed by exposure to low concentrations of gaseousethoxysilicates. This allows the removal of the hydrol-ysis product ethanol and prevents damage to the im-mobilized cells. Campostrini et al. immobilized plantcells on modified glass fibers. Sglavo et al. performedsimilar immobilizations upon collagen fibers [20].

The design requirements that have been identifiedby this previous research include 1) elimination of thecytotoxicity of the silica sol at gelation, which is dueprimarily to alcohol generation by the hydrolysis re-action and the presence of other organic species fora variety of reasons, and 2) precision control of theporous microstructure to provide a matrix that encour-ages colonization and mass transport. The productionroute described in this report attempts to meet these de-sign requirements by eliminating organic species from

the prehydrolyzed sol through distillation and control-ling pore size distribution through the introduction ofwater-soluble polymers into the gel.

Organic solvents are a common feature of almostall sol-gel production routes (including those that havebeen used for cell immobilization), and their elimina-tion can significantly influence the properties of the gel.These organic solvents often serve multiple purposesin the production of sol-gel silica, including decreasingthe polarity of the reaction solution to enable solvationof the alkoxy silicate precursor [21], acting as a volume“place holder” to enable the production of gels withsufficiently low density [22], providing a source of re-actant for the reverse silica solvation reaction duringaging [23], controlling drying during the production ofoptical-quality xerogels [24, 25], and acting as a polarphase in phase separation techniques [26]. The organicsolvents are most commonly the alcohol correspondingto the alkoxy substituents on the silicate precursor, butother solvents have been chosen and often yield uniquemicrostructural features [22, 24]. However, high con-centrations of organic solvents are toxic to biologicalsystems and limit the utility of these production routeswhere biocompatibility at gelation is an issue.

The current production route has been developedto yield sol-gel materials where essentially the onlysolvent present at gelation is water. In order to pro-duce sufficiently porous gels in the lack of an organic“place holder,” the hydrolysis ratios (molar ratio ofwater to alkoxy silicate) investigated here are severaltimes higher than in common sol-gel recipes and rangebetween 25 and 50. To the best of our knowledge,such ultra-high hydrolysis ratio sols have only beenthe subject of limited investigations [21], and severalresearchers have failed to obtain quality silica gels ateven lower hydrolysis ratios. The primary reason forthis failure appears to be difficulty in obtaining suf-ficient solubility of the alkoxy silicate in the aqueousphase. This issue has been addressed by the current pro-duction scheme by hydrolyzing the alkoxy silicate in alow pH aqueous solution until it is sufficiently polar tocompletely dissolve in the aqueous solvent. Once disso-lution has occurred, the prehydrolyzed sol is amenableto further manipulation to improve compatibility withbiological systems and manipulate microstructure forspecific applications. The prehydrolyzed sols describedhere were all distilled to remove as much hydrolysisreaction product alcohol as possible prior to the addi-tion of biological species. Other researchers have in-dicated that elevated temperatures during hydrolysis

Cells in Sol-Gels 271

may provide the additional advantage of more pre-cise control over the nature of the hydrolysis products[22, 27].

The second design issue is precision control ofthe porous microstructure to provide a matrix thatencourages colonization and mass transport yet pre-vents contamination by the external environment. AsPope pointed out, the mean pore diameter of commonsol-gel-derived silica gels is two orders of magnitudesmaller than the dimensions of common cells [12]. Thishas been shown to limit colonization of sol-gel-derivedsilica by entrapped microorganisms [19], and other re-searchers have shown that mass transport through suchgels is painfully slow [28]. Furthermore, optimal cellgrowth in ceramics has been achieved with pore diam-eters between one and four to five times the size of thecells [29, 30]. Clearly, a need exists for the productionof controlled macroporous gels for cell immobilization.

The production route described in this report pro-vides a simple and cytocompatible method to produceand engineer macroporous gels through the introduc-tion of water-soluble polymers to the prehydrolyzedsol. Water-soluble polymers have previously been usedin non-cytocompatible sol-gel routes to produce macro-porous samples by inducing a phase separation dur-ing gelation [31–34], but the resulting gels are oftenirreproducible and often do not produce large mono-lithic samples. In the current production route, the mi-crostructure of the final gel is reproducibly tunable andenables the production of monolithic macroporous gelsand aerogels.

Materials and Methods

Prehydrolyzed Sol Production

Ultra-high hydrolysis ratio prehydrolyzed sols wereprepared by heating a 100 : 0.1 molar ratio of distilledwater : nitric acid solution to the distillation tempera-ture (either 60◦C or 85◦C), followed by adding a suf-ficient volume of chilled (4◦C) tetraethylorthosilicate(TEOS, Aldrich Chem.) to obtain hydrolysis ratios of25, 33, or 50. This mixture was initially turbid andstirred at 800 RPM for 10 minutes. This was usuallysufficient to yield a transparent sol that was stirred at400 RPM for the remainder of the distillation. An openpot distillation was performed for 1 hr at 85◦C or 18 hrsat 60◦C. The solution was allowed to return to roomtemperature, at which time polyethylene glycol (Car-bowax Sentry Polyethylene Glycol 3350, DuPont) was

added in amounts between 0 and 2.5 grams per 100 mlsolution. The polyethylene glycol (PEG) was dissolvedthrough vigorous shaking.

Test Gel Production

Gelation was induced by raising the pH of the acidicsol to form so-called “two-step” gels at room tem-perature immediately after solvation of the polymer.Furthermore, the influence of time and temperature atgelation was investigated. Temperature effects were in-vestigated by heating or cooling aliquots of the sameprehydrolyzed sol/PEG mix to the desired temperaturebefore proceeding with gelation. Time effects were in-vestigated by allowing the distilled, prehydrolyzed solto “age” by stirring at room temperature in the absenceof PEG. When the desired time had been reached, PEGwas added to an aliquot of the sol before proceedingwith gelation.

For the majority of the results presented here, theexperimental focus was the reproducible productionand characterization of these novel, high hydrolysisratio gels rather than handling biological systems.Molar equivalent amounts (to nitric acid) of 1 M aque-ous potassium hydroxide solution were added to thepolymer-containing, prehydrolyzed sols. This inducedgelation prior to the addition of either cells or growthmedia to the sol. This was done because growth me-dia and cell solution composition is largely unknownand may vary or contain other constituents that influ-ence the final microstructure of the gel. Gel times werecommonly on the order of one minute and decreasedwith increasing polymer concentration. All gels wereaged in their production solution for four days at roomtemperature and displayed significant syneresis.

Many of the gel characterization techniques avail-able in our laboratory (nitrogen sorption, SEM) requiredry samples. In order to preserve as much of the mi-crostructure of the gels as possible, the gels were su-percritically dried to form aerogels. Aging in produc-tion solution was followed by three successive ethanolexchanges over a total of six days, and then yet an-other exchange with liquid carbon dioxide followed bysupercritical drying to obtain the final aerogel samples.

Gel Characterization

Immediately after supercritical drying, the gels wereweighed and the geometric dimensions measuredmanually. After an 18 hr evacuated bake at 250◦C, the

272 Conroy et al.

mass of the gels was again measured and these valueswere assumed to correspond to an approximate densitybefore and after removal of residual volatile contami-nants from the gels.

Aerogel samples were characterized through ni-trogen sorption porosimetry, scanning electron mi-croscopy (SEM), UV-vis spectroscopy, and acousticvelocity measurements. Nitrogen sorption porosime-try was performed using a Micrometrics ASAP PoreSize Analyzer on approximately 0.1 gram aerogel sam-ples after an 18 hr evacuated bake at 250◦C. The BETsurface area, BJH adsorption cumulative pore volumeand average pore diameter were calculated from theadsorption isotherms. SEM micrographs were takenusing a JEOL JSM-35 after vacuum sputtering withapproximately 200 angstroms of gold. UV-vis spec-troscopy was performed on samples cast and aged instandard 10 mm/3 ml polystyrene cuvettes. After aging,the samples were decast from the cuvettes, subject toethanol exchange, supercritically dried, and then placedinto Suprasil 300 quartz cuvettes and probed usinga Hewlett-Packard H.-P. 8453 UV-vis Spectrophoto-meter. Due to shrinkage during aging and/or super-critical drying, the final pathlength through the aero-gel samples was 0.9011± 0.008 cm. Pulse transit timemeasurements of acoustic velocity were made on sam-ples under ambient conditions using a PanametricsPulser/Receiver 5055PR and ultrasonic preamplifierin conjunction with a matched pair of 5 MHz con-tact mode transducers (Panametrics, Model # V109).Coupling with the gel samples was performed using asingle Parafilm layer placed between the transducersand the gels. The strain variability that has previouslybeen observed in gel samples [35, 36] was accomo-dated by measuring at the lowest possible stress thatyielded an apparent signal.

Cell Immobilization

Gelation can be induced by addition of a cytocompati-ble amount of (buffered) growth medium powder to theprehydrolyzed sol, followed by addition of the biologi-cal system of interest. In this study Luria-Bertani brothpowder (Difco), containing yeast extract (5 g/L), NaCl(10 g/L) and tryptone (10 g/L), was added at the manu-facturers recommended concentration of 2.5 g/100 mlto the prehydrolized sol. The powder was dissolved byrapid shaking, 0.6 ml of 1 M KOH wasadded to bringthe pH of the sol up to 6.0, and 0.5 ml ofE.colibacteriaper 20 ml sol was immediately added. The bacteria

had been grown to mid-logarithmic phase, which cor-responded to a concentration of 1013 cells/ml. Viabilityof the bacteria within the resulting gel was determinedmicroscopically using a LIVE/DEAD Baclight bacteriaviability kit (Molecular Probes, Eugene, OR) followingthe manufacturer’s instructions.

Results and Discussion

Characterization of the gels was performed using anumber of different analytical techniques. SEM mi-croscopy provided clear images of macrostructural fea-tures, whereas nitrogen sorption was used to probethe mesopore regime. UV-vis transmittance measure-ments probed features in both the macro- and meso-pore regime. Acoustic velocity measurements provideda very sensitive probe of gel features, although corre-lating velocity measurements with discrete microstruc-tural features proved difficult.

Pore Size as a Function of PolyethyleneGlycol Concentration

The most obvious and perhaps most important conclu-sion that can be drawn from the various analytical tech-niques is that increasing polyethylene concentrationyields macroporous sol-gels with increasingly largepores and/or backbone structures. The UV-vis trans-mission measurements given in Fig. 1 clearly showthat the transmittance range of the gels decreases withincreasing polymer concentration.

Figure 1. UV-vis transmittance of various aerogels showing theinfluence of PEG concentration. The gels were cast into optical cu-vettes immediately after a one hour distillation at 85◦C and had ahydrolysis ratio of 33. The decrease in transmissivity with increas-ing PEG concentration is presumably due to increased feature sizeand scattering.

Cells in Sol-Gels 273

The nature of the microstructure that causes the in-creased scattering is more difficult to quantify. For agiven hydrolysis ratio and distillation time, the den-sity of the final aerogel samples is identical, within thelimits of our measurement techniques. If it is assumedthat any change in density associated with changes insurface area (due to, for example, extent of surfacesilanol condensation) is negligible, then the porosity ofthe gels with a given hydrolysis ratio should be identi-cal. However, the cumulative pore volume observedthrough nitrogen sorption measurements clearly de-creases (Fig. 2(b)) with increasing PEG concentration.

The apparent decrease in porosity observable withnitrogen sorption must then correspond to a concomi-tant increase in porosity outside of the mesopore range.

(a) (b)

(c)

Figure 2. (a)–(c): Calculated (a) BET surface area, (b) BJH adsorption cumulative pore volume, and (c) BJH adsorption mean pore diameteras a function of PEG concentration. The measured values of all three decrease with increasing PEG concentration, regardless of distillationconditions. Measured pore volume (b) and mean pore diameter (c) decrease with increasing PEG concentration. However, since the density ofthe gels is nearly identical, this probably represents an increased fraction of the porosity shifting outside the mesopore range. A longer and lowertemperature distillation appears to produces gels with an increased fraction of the porosity still within the mesopore range.

The nitrogen sorption isotherms in Figs. 3 and 12display slight decreases in adsorption at the lowestmeasured pressures. This may correspond to slight de-creases in microporosity of the aerogel samples, or itmay be a result of shifting the relatively polydispersepore distribution to larger pore sizes and away from themicropore regime. Indeed, the shift to large pore sizeswith increasing PEG concentration seen in Figs. 3 and12 is so dramatic that the highest PEG concentrationsyield isotherms that approach Type I behavior. Thiscan be seen most dramatically in Fig. 3, where the fi-nal isotherm adsorbs less than 400 cm3/g at a relativepressure of 0.8.

The macroporous structures resulting from increa-sed PEG concentration are readily visible under

274 Conroy et al.

Figure 3. Nitrogen sorption isotherms of various aerogels showingthe influence of PEG concentration. The aerogels had a hydrolysisratio of 33 and were gelled after an 18 hour distillation at 60◦C. Aspolyethylene glycol concentration increases, the amount of nitrogensorption decreases. It is believed that this is a consequence of anincreased fraction of the porosity shifting outside the mesopore range.

scanning electron microscopy examination. For exam-ple, Fig. 4(a)–(c) clearly shows an increase in bothsolid structure size and pore size with increasing PEGconcentration. Because of the increase in solid struc-ture size, we suspect that the influence of PEG con-centration upon gel structure is related to the point atwhich phase separation occurs in the prehydrolyzedsols. If phase separation occurs sooner during gela-tion due to increased polymer concentration, growthof a uniformly-distributed solid silica network will becurtailed as low molecular weight species are con-fined to smaller fractions of the solid volume. Gelswith larger features and pore diameters will thusresult.

The control of microstructure provided by PEGconcentration is quite facile. Although the changesobserved by nitrogen sorption appear to become lesssignificant with increasing PEG concentration, this isprobably due to the shift of an increasing fraction ofthe porosity out of the range accessible through nitro-gen sorption. Even neglecting the shift of the poros-ity outside the mesopore regime, the mutability in mi-crostructure is quite high. For example, the decreasein BET surface area with PEG concentration is onthe order of 200 m2/g per gram PEG per 100 ml ofprehydrolyzed sol. Likewise, the decrease in cumula-tive pore volume and mean pore diameter accessiblewith nitrogen sorption is on the order of 11 m3 and30 nm per gram PEG per 100 ml of prehydrolyzed sol,respectively.

Figure 4. SEM micrographs at 10,000× magnification of variousaerogels formed after a 1 hrdistillation at 85◦C with a hydrolysisratio of 50. Samples (a)–(c) were formed at PEG concentrationsof 0.15, 0.35, and 0.6 w/v% respectively. In the macropore regimevisible here, the pore structure appears to transition from a “colloidal”assemblage of particles to a “polymeric” interconnected network.

Cells in Sol-Gels 275

Figure 5. Nitrogen sorption isotherms of various aerogels showingthe influence of “aging” the silica sol. The gels had a hydrolysis ratioof 33, a PEG concentration of 0.705 w/v%, and were distilled forone hour at 85◦C and cooled to room temperature before time wasset to zero. Neutralization of the acidic hydrolysis catalyst was per-formed at the times indicated and gelation proceeded immediately.In this particular grouping, as time progressed, the amount of nitro-gen adsorption increases. It is believed that this is a consequence ofincreased polycondensation of the prehydrolyzed sol with time.

Sol “Aging” Prior to Gelation

Despite the low pH of the distilled, prehydrolyzedsol, the polycondensation reaction of various silicicacid species will continue, leading eventually to gela-tion [37]. In order to investigate the influence of thiscontinued polycondensation upon gel micro- andmacro-structure, the second step of the gelation (pHneutralization) was performed at various times aftercompletion of the distillation. Nitrogen sorption iso-therms for a set of three gels with identical polyethy-lene glycol concentrations (0.705 g/ml) and hydrolysisratios (33) but neutralized at different times are shownin Fig. 5. Despite the near overlap of these isotherms,slight differences in gel structure were calculated fromthese isotherms. Figure 6(a)–(c) indicates slight in-creases in cumulative pore volume and mean pore di-ameter and both increases and decreases in surface areawith time for a variety of gel samples. Perhaps the mostimportant conclusion that can be drawn from this datais that the rate of change in the parameters measured bynitrogen sorption is very small. For example, if one as-sumes a linear increase of these parameters with time,then the rate of increase in surface area with time isbetween−2 and 6 m2/g/hr, the rate of increase in cu-mulative pore volume is between 0.005 and 0.04 m3/hr,and the rate of increase of mean pore diameter with timeis between 0.04 and 0.06 nm/hr. This indicates a fairlyrobust system where PEG concentration is a much more

(a)

(b)

(c)

Figure 6. (a)–(c): Calculated (a) BET surface area, (b) BJH ad-sorption cumulative pore volume, and (c) BJH adsorption mean porediameter as a function of time of gelation after completion of distil-lation. The influence of distillation time upon the apparent surfacearea (a) of the aerogels is difficult to identify over these times. Theapparent pore volume (b) and mean pore diameter (c) of the aerogelsappears to increase slightly with increasing sol “aging” time, re-gardless of PEG concentration. This may be related to the increasesapparent in Fig. 2 due to increased time of distillation.

276 Conroy et al.

significant determinant of microstructure than time ofgelation.

In the macropore regime, scanning electron mi-croscopy (Fig. 7) does not indicate major changes instructure with sol “aging” time for the same gels whoseadsorption isotherms are shown in Fig. 5. The gelsare essentially indistinguishable and indicative of therobustness of the process.

Acoustic velocity measurements (Fig. 8) provide theclearest indication of the influence of sol aging timeupon the structure of the final gels. The acoustic ve-locity of the various gels increases noticably with solaging time over the time periods examined. The rateof change in acoustic velocity appears to increase withtime. Over the initial 16 hr period examined, the rateof increase in velocity averages between 0.5 and 1.5m/s/hr. Since the density of these gels should remainconstant, the increase in acoustic velocity is presum-ably due to an increase in the modulus of the gel. Themoduli of these gels is in the range of 6.3 to 0.72 MPa.The decrease in acoustic velocity with increasing PEGconcentration is consistent with a Hertzian interpreta-tion of acoustic propagation through porous media. Inshort, as the gels become more “colloidal,” the diam-eter of the contact “necks” between adjacent particlesrelative to the the diameter of the particles themselvesdecreases. The majority of the deflection (at small de-flections) of a particle-neck-particle system occurs atthe neck, and thus the apparent modulus of the en-tire system is decreased relative to a comparable massallocated as, for example, a single cylinder. This will bediscussed more extensively for these sol-gel systems ina forthcoming publication.

It should be pointed out that the density of all the gelsin Fig. 8 is nearly identical and thus the longitudinalacoustic velocity through the gels appears nearly in-dependent of density. Previous researchers have at-tempted to relate density to acoustic velocity throughaerogels, and even proposed acoustic velocity mea-surements as an approach for determining density [36,38–41]. This is clearly not possible with these samples.

Although the changes in microstructure in the regimeexamined by nitrogen sorption appear small, they arepresumably due to an increased extent of “polymeric”polycondensation prior to addition of both the PEG andbase and will eventually have a significant influenceupon gel structure. Previous researchers have discussedsingle step, acid-catalyzed gels in terms of linear chaingrowth mechanisms that eventually lead to entangle-ment of “polymeric” condensates and high modulus,high surface area, low pore size gels. The increased

Figure 7. SEM micrographs at 2,000×magnification of variousaerogels formed with a hydrolysis ratio of 33 and a PEG concentra-tion of 0.705 w/v%. Samples (a)–(c) were gelled 0, 18, and 25 hrsafter a 1 hrdistillation at 85◦C, respectively. The influence of agingtime apparent from these micrographs is minimal.

Cells in Sol-Gels 277

Figure 8. Longitudinal acoustic velocity in various aerogels as afunction of time of gelation after distillation. Interestingly, acousticvelocity measurements provide the clearest indication of the influ-ence of sol “aging” time. The increased acoustic velocity may berelated to increased polycondensation of the prehydrolyzed sol un-der acidic conditions, as gels formed under such conditions tend topossess higher elastic moduli.

degree of condensation in the sol at the time of thesecond step neutralization may lead to a less compactsilica backbone structure with a larger pore volumewithin the range measurable by nitrogen sorption. Astime progresses, these effects are expected to becomemagnified and to eventually influence macroporestructure as well.

Temperature Measurements

The temperature of the prehydrolyzed sol at gelationis an issue in bioencapsulation for several reasons.Several researchers have cooled prehydrolyzed sols inorder to slow gelation after neutralization in two-stepsynthetic routes. Furthermore, temperature will havea profound impact upon metabolism of the cells ormicroorganisms to be immobilized. Temperature alsohas a significant influence upon the macrostructure ofphase-separated, macroporous sol-gel synthetic routes.For these reasons, the influence of temperature upon gelmicrostructure in the current route was investigated.

Visual examination of gels produced at 8, 16, 22,34, and 43◦C indicates significant differences betweengels. As gelation temperature increases, the gels be-come more opaque and macroscopic heterogeneitiesbecome apparent to the naked eye at the highest tem-peratures. Nitrogen sorption measurements however

Figure 9. Nitrogen sorption isotherms of various aerogels showingthe influence of temperature at gelation upon structure. The gels hada hydrolysis ratio of 33, a PEG concentration of 0.5 w/v%, and weredistilled for one hour at 85◦C. Neutralization of the acidic hydroly-sis catalyst was performed at the temperatures indicated. Althoughthe differences between the gels were macroscopically apparent, theadsorption isotherms are nearly indistinguishable from one another.

indicate that the majority of these changes occur onlywithin the macroscale. The adsorption isotherms arenearly identical for all temperatures examined (Fig. 9).The calculated BET surface area indicates a slight de-crease in surface area with increasing temperature,on the order of 4 m2/g◦C (Fig. 10(a)). The cumula-tive pore volume (Fig. 10(b)) and mean pore diameter(Fig. 10(c)) are also remarkably robust, decreasing by0.006 cm3/g◦C and increasing by 0.014 nm/◦C, respec-tively. These changes are on the order of magnitude ofthe changes due to aging time of the prehydrolyzed sol,and thus are still much smaller than the changes due topolyethylene glycol concentration.

Acoustic velocity measurements also fail to providea clear indication of significant differences between thegels (Fig. 11). Acoustic velocity appears to be relativelyindependent of temperature in the range examined.Although there are large heterogeneities apparent tothe naked eye in the monolithic gels, they appear tohave a minimal influence upon acoustic velocity.

In addition to the uncertainty in the magnitude ofthe temperature effects upon gel structure, identifica-tion of the origin of any temperature dependency isalso difficult. The polycondensation rate will be a func-tion of temperature, as will the phase separation behav-ior of the polyethylene glycol/silicic acid sol system.The interactions between these various effects are dif-ficult to elucidate and will be the subject of furtherstudies.

278 Conroy et al.

(a)

(b)

(c)

Figure 10. (a)–(c): Calculated (a) BET surface area, (b) BJH ad-sorption cumulative pore volume, and (c) BJH adsorption mean porediameter as a function of temperature at gelation. The apparent sur-face area (a) and pore volume (b) of the aerogels may decreaseslightly with increasing temperature, whereas mean pore diameterappears to be independent of temperature. The origin of these effectsis unknown, as both the condensation reaction and phase separationbehavior of the system will be influenced by temperature. Neverthe-less, these effects appear minimal relative to the influence of PEGconcentration and are indicative of the robustness of the productionroute.

Figure 11. Longitudinal acoustic velocity as a function of tempera-ture at gelation. The acoustic velocity also appears to be independentof temperature. The standard deviation of the velocity measurementsis smaller than the marker size.

Distillation Conditions

The influence of distillation conditions upon gel mi-crostructure was investigated solely through the useof nitrogen sorption porosimetry. Figure 2(a)–(c) com-pares BET surface area, cumulative pore volume, andmean pore diameter for several different polyethyleneglycol concentrations for both an 18 hr distillation at60◦C and a one hour distillation at 85◦C. The isothermsfrom which this data are derived is respectively shownin Figs. 3 and 12. The longer distillation appears to pro-duce gels with decreased surface area, increased meanpore diameter, and a slight decrease in cumulative porevolume (in the range measured by nitrogen sorption).

Figure 12. Nitrogen sorption isotherms of various aerogels show-ing the influence of PEG concentration. The aerogels had a hydrolysisratio of 33 and were gelled after a one hour distillation at 85◦C. Aspolythylene glycol concentration increases, the amount of nitrogensorption decreases. It is believed that this is a consequence of an in-creased fraction of the porosity shifting outside the mesopore range.

Cells in Sol-Gels 279

One can choose to examine the influence of distilla-tion conditions in light of the previously described sol“aging” data. The distillation can be viewed as anotheraging experiment at elevated temperatures. For exam-ple, the elevated pore volume and mean pore diameterafter an 18 hr distillation may arise from the same ef-fects that cause increases in these parameters duringaging. If this is the case, then the clear decrease in sur-face area apparent from the distillation data may indi-cate that the surface area in the aging experiments willlikewise begin to decline as time progresses beyondthe scope of the data shown in Fig. 2(a)–(c). This sortof analysis should, however, only be cautiously pur-sued, as others have shown that the equilibrium formsof polycondensed silicic acid are dependent upontemperature within the range examined here [27].

Hydrolysis Ratio

The influence of hydrolysis ratio upon gel structure wasperhaps the most difficult parameter to examine duringthe course of these investigations. The hydrolysis ra-tio had a significant influence upon the foaming andgelation behavior of the prehydrolyzed sol. In general,the most uniform and robust gels were obtained at hy-drolysis ratios of 33 and PEG concentrations rangingbetween approximately 0.03 and 0.4 g per 20 ml sol.At lower PEG concentrations, the gels tended to crackupon supercritical drying, while higher PEG concen-trations tended to foam during mixing and/or flocculateprior to gelation. Furthermore, the gels themselves be-came very friable at the highest PEG concentrations.

Figure 13. Mean density and standard deviation of the dried aero-gels as a function of hydrolysis ratio of the gels. As hydrolysis ra-tio increases, density decreases. With water the only volume “placeholder” in gels, the decrease in density is due to the lower volumefraction of silica in the sol.

(a)

(b)

(c)

Figure 14. (a)–(c): Calculated (a) BET surface area, (b) BJH ad-sorption cumulative pore volume, and (c) BJH adsorption mean porediameter as a function of PEG concentration. It appears that the sur-face areas of the lowest (25) and highest (50) examined hydrolysisratios are more similar to each other than they are to the intermediatehydrolysis ratio (33). This may be due to the intermediate hydrolysisratio being closer to the critical concentration for this system.

280 Conroy et al.

Data has been collected with some gels formed at thesehigh PEG concentrations, although these should beviewed skeptically in light of possible inhomogeneitiesin the final gels. The other hydrolysis ratios (25 and 50)that were examined were also less successful than theintermediate hydrolysis ration of 33. Higher hydrolysisratio gels (50) were more prone to cracking duringhandling and after drying and lower hydrolysis ratiogels (25) tended to gel almost immediately upon addi-tion of base and hence displayed large inhomogeneities.Rapid gelation and foaming led to macroscopic hetero-geneities and great difficulty in analyzing the gels. Theinfluence of the nature and chain length of the polymerphase will be the subject of future investigations, al-though preliminary studies with a PEG/polypropylenecopolymer indicate that foaming occurs at lower poly-mer concentrations than with the polymer used in thesestudies.

The most apparent influence of stoichiometry upongel structure was in the density of the final gel. Asshown in Fig. 13, density decreases with increasing hy-drolysis ratio. This is presumably due to the decreasedfraction of silica precursor in the prehydrolyzed sol.It is problematic to calculate the percent of the initialTEOS that is found in the final gels because of thepossible inclusion of unhydrolyzed and/or re-esterifiedsurface sites.

Despite the difficulty in handing, gels formed atdifferent hydrolysis ratios were examined using nitro-gen sorption porosimetry (Fig. 14(a)–(c)) and scanningelectron microscopy (Fig. 15). Interestingly enough,samples with hydrolysis ratios of 25 and 50 appearmore similar to one another than samples formed withhydrolysis ratios of 33. In both cases, gels formed withsols at hydrolysis ratios of 25 and 50 display both adecreasedandmore rapidly decreasing with increas-ing PEG concentrationsurface area, cumulative porevolume, and mean pore diameter relative to sols formedwith hydrolysis ratios of 33. These effects are difficultto quantify from SEM micrographs. However, the gelsin Fig. 15 may appear more globular or colloidal withincreasing hydrolysis ratio. The relationship betweenthe qualitative appearance of the gels and the nitrogensorption data is unclear.

The similarity between the higher and the lowerhydrolysis ratio samples may be related to the phaseseparation responsible for the formation of thesemacroscopic gels. In terms of a Flory-Huggins quasitwo phase system (which has previously been appliedto silica sol-gel systems [34]), the hydrolysis ratios may

Figure 15. SEM micrographs at 2,000×magnification of variousaerogels formed with PEG concentrations of 0.75± 0.05 w/v% aftera 1 hr distillation at 85◦C. Samples (a)–(c) had hydrolysis ratios of 50,33, and 25, respectively. The influence of hydrolysis ratio apparentfrom these micrographs is minimal.

Cells in Sol-Gels 281

Figure 16. Fluorescent light micrographs at 200×magnification ofE.coli 24 hr after immobilization in a 0.5% PEG gel produced by themethod described above, using theBacLightlive/dead staining kit. (a) Live cells (labeled with green fluorescence). (b) Dead cells (arrow; labeledwith red fluorescence). The micrographs were taken at a single location with two filter sets indicating that the vast majority of the bacteria wereviable.

282 Conroy et al.

be above and below the critical concentration for thesilicic acid condensation species in the water/PEG so-lution. It is difficult to draw more detailed conclusionsfrom these preliminary results, and further investiga-tions into the influence of hydrolysis ratio upon thestructure of these gels will be performed.

Cell Immobilization

One example of the cell immobilization that is possibleusing these gels is shown in Fig. 16. This fluorescentlight micrograph at 200X magnification demonstratesthe extent of survival ofE.coli 24 h after immobiliza-tion in a gel. This micrograph indicates that a largenumber of bacteria survived the gelation as indicatedby the density of live cells (16(a)). Additionally, thenumber of dead cells present (16(b)) was very low incomparison. Although only dead cells that remain in-tact are detected by this method these results do suggestthat a large proportion of the bacteria remain viable.When bacteria were added to a traditional TEOS/watersol, with no distillation or ionic compensation, no cellswere observed in the resulting gels, indicating that allthe cells were lysed. Future reports will detail both theshort and long term survival rate of a variety of immobi-lized cells in the different matrices accessible throughthe route described here.

Conclusion

Sol-gel materials have long held promise as immo-bilization matrices. By working in an aqueous pre-hydrolyzed sol that contains essentially only prehy-drolyzed precursors to silica, the cytocompatability ofthe sol-gel process should be dramatically improved.Furthermore, water-soluble polyethylene glycol can beadded to this prehydrolyzed sol to produce macrop-orous gels that may be amenable to colonization andfacile mass transport.

The outlined production route is both easy androbust. Many of the parameters that were suspectedto have a significant influence upon structure of thefinal gel were shown to have minimal and/or an un-measurably small influence. Neither temperature, sol“aging” time, nor distillation conditions were found toexert as strong an influence upon gel microstructure asPEG concentration. This is itself unusual: many cur-rent macroporous production routes are unreliable andincapable of forming large monolithic aerogels.

Future work will include examining the effects ofinherently poorly characterized and non-uniform cellgrowth media and of the cells themselves upon the mi-crostructure and properties of sol-gels produced by thisroute. Given the lability of the sol-gel reactions, theseelements are anticipated to have a significant influence.For example, simply increasing the ionic strength ofthe sol will likely increase the rate of condensation ascharge screening occurs. Furthermore, bacterial growthmedia while largely undefined is known to contain bothpolypeptides and carbohydrates that act as dispersantsin other colloidal systems. Through the further devel-opment of this sol-gel production route, we hope todevelop gels that are useful in a variety of tissue engi-neering, bioreactor, and biosensor applications.

Acknowledgments

The authors would like to thank Christina Elzey forher assistance with the scanning electron microscopy,Catherine Iliesa and Emilie Ariane for their editorialcomments, and Mungo Marsden for providing the bac-teria. This work was sponsored by DARPA undercontract number MDA972-97-C-0020.

References

1. S.I. Anderson, S. Downes, C.C. Perry, and A.M. Caballero,J. Mat. Sci.: Mat. Med.9, 731 (1998).

2. K. Tsuru, C. Ohtsuki, A. Osaka, T. Iwamoto, and J.D.Mackenzie, J. Mat. Sci.: Mat. Med.8, 157 (1997).

3. S.-B. Cho, K. Nakanishi, T. Kokubo, N. Soga, C. Ohtsuki, andT. Nakamura, J. Biomed. Mat. Res.33, 141 (1996).

4. S.-B. Cho, F. Miyaji, T. Kokubo, K. Nakanishi, N. Soga, and T.Nakamura, J. Biomed. Mat. Res.32, 375 (1996).

5. V. Gross, R. Kline, H.J. Schmitz, and V. Strunz, CRC Crit. Rev.Biocomp.4, 2 (1988).

6. L.L. Hench, inBioceramics, Vol. 523:Materials Characteristicsversus In-Vivo Behavior, edited by P. Ducheyne and J. Lemmons(Ann. New York Acad. Sci., 1988), p. 54.

7. M. Uo, K. Yamashita, M. Suzuki, E. Tamiya, I. Karube, andA. Makishima, J. Ceram. Soc. Jpn.100, 426 (1992).

8. G. Carturan, R. Campostrini, S. Dire, A. Scardi, and E. DeAlterius, J. Mol. Catal.57, L13 (1989).

9. E.J.A. Pope, J. Sol-Gel Sci. Tech.4, 225 (1995).10. M. Al-Saraj, M.S. Abdel-Latif, I. El-Nahal, and R. Baraka,

J. Non-Cryst. Sol.248, 137 (1999).11. T. Hino, H. Yamada, and S. Okamura, US Patent 4,148,689,

1979.12. E.J.A. Pope, K. Braun, M. Van Hirtum, C.M. Peterson, P. Tresco,

and J.D. Andrade, inSol-Gel Science and Technology, editedby E.J.A. Pope, S. Sakka, and L. Klein (Am. Ceramic Soc.,Westerville, Ohio, 1995). Ceramic Transactions, Vol. 55, p. 33.

13. E.J.A. Pope, US Patent 5,693,513, 1997.

Cells in Sol-Gels 283

14. E.J.A. Pope, US Patent 5,739,020, 1998.15. E.J.A. Pope, K. Braun, and C.M. Peterson, J. Sol-Gel Sci. Tech.

8, 635 (1997).16. K.P. Peterson, C.M. Peterson, and E.J.A. Pope, Proc. Soc. Exp.

Bio. Med.218, 365 (1998).17. E.J.A. Pope, US Patent 5,895,757, 1997.18. J. Livage, J.Y. Barreau, J.M. Da Costa, and I. Desportes, SPIE

Proc. Ser. (Sol-Gel Optics III)2288, 493 (1994).19. T. Branyik, G. Kuncova, J. Paca, and K. Demnerova, J. Sol-Gel

Sci. Tech.13, 283 (1998).20. V.M. Sglavo, G. Carturan, R. Dal Monte, and M. Muraca, J. Mat.

Sci.34, 3587 (1999).21. L.C. Klein and G.J. Garvey, inBetter Ceramics Through Chem-

istry, edited by C.J. Brinker, D.E. Clark, and D.R. Ulrich (North-Holland, New York, 1984). MRS Symp. Proc., Vol. 32, p. 33.

22. T.M. Tillotson, J.F. Poco, L.W. Hrubesch, and I.M. Thomas, USPatent 5,409,683, 1995.

23. J.K. West, R. Nikles, and G. Latorre, inBetter Ceramics ThroughChemistry, edited by C.J. Brinker, D.E. Clark, and D.R. Ulrich(North-Holland, New York, 1988). MRS Symp. Proc., Vol. 121,p. 219.

24. L.L. Hench, G. Orcel, and J.L. Nogues, inBetter CeramicsThrough Chemistry, edited by C.J. Brinker, D.E. Clark, and D.R.Ulrich (North-Holland, New York, 1986). MRS Symp. Proc.,Vol. 73, p. 35.

25. L.L. Hench, inBetter Ceramics Through Chemistry, edited byC.J. Brinker, D.E. Clark, and D.R. Ulrich (North-Holland, NewYork, 1984). MRS Symp. Proc., Vol. 32, p. 101.

26. K. Tadanaga, K. Iwashita, T. Minami, and N. Tohge, J. Sol-GelSci. Tech.6, 107 (1996).

27. B. Unger, H. Jancke, M. Haehnert, and H. Stade, J. Sol-Gel Sci.Tech.2, 51 (1994).

28. B. Hosticka, P.M. Norris, J. Brenizer, and C. Daitch, J. Non-Cryst. Solids225, 293 (1998).

29. R.A. Messing and R.A. Opperman, Biotech. and Bioeng.21, 49(1979).

30. K. Iwasaki and N. Ueno, J. Ceram. Soc. Jpn. Int. Ed.98, 13(1990).

31. K. Kajihara, K. Nakanishi, and K. Tanaka, J. Am. Ceram. Soc.81, 2670 (1998).

32. H. Minakuchi, K. Nakanishi, and N. Soga, Anal. Chem.68, 3498(1996).

33. K. Nakanishi, N. Soga, and H. Matsuoka, J. Am. Ceram. Soc.75, 971 (1992).

34. H. Kaji, K. Nakanishi, and N. Soga, J. Sol-Gel Sci. Tech.1, 35(1993).

35. J.F.T. Conroy, B. Hosticka, S.C. Davis, A.N. Smith, and P.M.Norris, Microscale Thermophys. Engin.3, 199 (1999).

36. H. Altmann, T. Schlief, J. Gross, and J. Fricke, Ultrasonic Inter-national 91 Conf. Proc. 261 (1991).

37. K.R. Andersson, L.S.D. Glasser, and D.N. Smith, inSolubleSilicates, edited by J.S. Falcone (American Chem. Soc.Washington, D.C., 1982). ACS Symp. Ser., Vol. 194, p. 115.

38. J. Gross, J. Fricke, and L.W. Hrubesh, J. Acoust. Soc. Am.91,2004 (1992).

39. J. Gross and J. Fricke, J. Non-Cryst. Sol.145, 217 (1992).40. J. Gross, J. Fricke, R.W. Pekala, and L.W. Hrubesh, Phys. Rev.

B 45, 12774 (1992).41. J. Gross, G. Reichenauer, and J. Fricke, J. of Phys. D: Appl.

Phys.91, 1447 (1988).